Performance?

Yes, we did monitor performance. But it simply was not worth talking about: the results at 20°C inlet are almost identical to those at 40°C inlet. The only difference that lower temperatures could make is a slight increase in the amount of time spent at higher Turbo Boost frequencies, but we could not measure any significant difference. The reason is of course that some of our VMs are also somewhat disk intensive.

Conclusion

The PUE optimized servers can sustain up to 40°C inlet temperature without a tangible increase in power consumption. It may not seem spectacular but it definitely is. The "PUE optimized" servers are simply improved versions; they do not need any expensive technology to sustain high inlet temperatures. As a result, the Supermicro Superserver 6027R-73DARF cost is around $1300.

That means that even an older data center can save a massive amount of money by simply making sure that some sections only contain servers that can cope with higher inlet temperatures. An investment in air-side or water-side economizers could result in very large OPEX savings.

Reliability was beyond the scope of this article and the budget of our lab. But previous studies, for example by IBM and Google, have also shown that reasonably high inlet temperatures (lower than 40°C) have no significant effect on the reliability of the electronics.

Modern data centers should avoid servers that cannot cope with higher inlet temperature at all cost as the cost savings of free cooling range from significant to enormous. We quote a study done on a real-world data center by Intel:

"67% estimated power savings using the (air) economizer 91% of the time—an estimated annual savings of approximately USD 2.87 million in a 10MW data center"

A simple, solid and very affordable server without frills that allows you to lower the cooling costs is a very good deal. 

Other Components
Comments Locked

48 Comments

View All Comments

  • extide - Tuesday, February 11, 2014 - link

    Yeah there is a lot of movement in this these days, but the hard part of doing this is at the low voltages used in servers <=24v, you need a massive amount of current to feed several racks of servers, so you need massive power bars and of course you can lose a lot of efficiency on that side as well.
  • drexnx - Tuesday, February 11, 2014 - link

    afaik, the Delta DC stuff is all 48v, so a lot of the old telecom CO stuff is already tailor-made for use there.

    but yes, you get to see some pretty amazing buswork as a result!
  • Ikefu - Tuesday, February 11, 2014 - link

    Microsoft is building a massive data center in my home state just outside Cheyenne, WY. I wonder why more companies haven't done this yet? Its very dry and days above 90F are few and far between in the summer. Seems like an easy cooling solution versus all the data centers in places like Dallas.
  • rrinker - Tuesday, February 11, 2014 - link

    Building in the cooler climes is great - but you also need the networking infrastructure to support said big data center. Heck for free cooling, build the data centers in the far frozen reaches of Northern Canada, or in Antarctica. Only, how will you get the data to the data center?
  • Ikefu - Tuesday, February 11, 2014 - link

    Its actually right along the I-80 corridor that connects Chicago and San Francisco. Several major backbones run along that route and its why many mega data centers in Iowa are also built along I-80. Microsoft and the NCAR Yellowstone super computer are there so the large pipe is definitely accessible.
  • darking - Tuesday, February 11, 2014 - link

    We've used free cooling in our small datacenter since 2007. Its very effective from september to april here in Denmark.
  • beginner99 - Tuesday, February 11, 2014 - link

    That map from Europe is certainly plain wrong. Especially in Spain btu also Greece and italy easily have some day above 35. It also happens couple of days per year were I live, a lot more north than any of those.
  • ShieTar - Thursday, February 13, 2014 - link

    Do you really get 35°C, in the shade, outside, for more than 260 hours a year? I'm sure it happens for a few hours a day in the two hottest months, but the map does cap out at 8500 out of 8760 hours.
  • juhatus - Tuesday, February 11, 2014 - link

    What about wear&tear at running the equipment at hotter temperatures? I remember seeing the chart where higher temperature = shorter life span. I would imagine the OEM's have engineered a bit over this and warranties aside, it should be basic physics?
  • zodiacfml - Wednesday, February 12, 2014 - link

    You just need constant temperature and equipment that works at that temperature. Wear and tear happens significantly at temperature changes.

Log in

Don't have an account? Sign up now